millenniumwaraigisfandomcom-20200223-history
Talk:Holy Warrior's Challenge/@comment-30176423-20161116032002/@comment-212.67.156.142-20161116182613
@Ssvb My theory is that people are unconsciousnessly p-hacking - both random people and people trying to more actively 'guess out the RNG'. Changing the lenght of the experiments when stuff doesnt happen? Taking data that only fits your theory? 'Testing' dozen of theories at the same time (possibly A LOT as there is a lot of RNG in the game - combination (CR/SU), drop rates, shrine pulls and more), that practically mean that they do hundreds experimentss at the same time, and go apeshit when few of them do not fit their human idea of RNG (and humans are super bad RNG generators, as there are too much patterns in them). When you are trying to find the corelations between some-random-data including your party, hour, how many times you rolled in a row etc. etc. and drop rate then at small-to-medicore sample size (and/or low RNG propabilities on top of that) and you WILL find some kind of correlation to brag about. I belive that my theory is pretty sound as well (as this exact behavior wrecks a lot of scientific research that are done 'to proof wild stuff' without actually 'having to really proof stuff'. There were papers that 'have proven' that humans can predict the future. How? They asked people to chose one of two options on PC screen, and then showed them what they picked. What happened? Researchers found that people pressed sexy-stuff more often than scary-stuff, using pretty small but scientifically acceptable sample size. (and I hope I dont have to convince you that people can predict the future when clicking at place when random tagged photos pop out) How to do such a research? Ez. Make a lot of groups. A lot and lot of them, give everyone a lot of attributes. Time of the day, their age, their gender, make light in the room vary from person to person and grab all data. Then have as much theories to test as possible. Have photos be tagged in many many ways. Have one photo have a bunch of tags. Scary stuff, sexy stuff, dark photos, still life, amount of green on the photo - whatever you will be able to grab the data about, the more the better. Then do reseach, and as you do it, search for correlations. You have X*Y theories that you can try to proof if you get right data. You can prove that people click sexy stuff more often, or that they will make the choice faster, or whatever. The more theories you test, the bigger the chance to get one is. And the last part of p-hacking - do only enough test for you to be considered of big enough sample size. 600 tests and still no theory? do the 100 more and test them again! Many of them can be just random fluctuations, but if you get lucky and get a lot of answers swinging you in the right direction, do just enough tests to be on '52%' or whatever you are required to meet to post your paper. DO NOT DO ANY MORE RESEARCH - if it was just a random thing it can ruin your 52% needed for your thesis. You can also try to write-off data that are not good for your thesis, and try to undermine them. What is the most interesting, I find all these behaviors in people trying to proove their 'RNG-hacks' here. Changing your party untill you 'figure out the composition' (or rather get lucky)? Check. Test all the theories at the same time? Check. Have the bias to theories that prove stuff over the data that showed that there are no correlation (have only pro-correlation data published and spoken about)? Check. I see no reason for drops to be random, I see no good implementation reason either, and I see no data from JP version that confirm these ideas (we sometimes have the differences like 50% and 52% drop rate, but IMO so small differences could be the effect of some floating point calculations, having drop rate be underrepresented or overrepresented, but this would be constant for every player, and not change the outcome of the game (and cannot be 'hacked' either - it is just 127,6 being rounded-up to 128 kind of deal - sure it is, its cool to get more rather than less, but nothing more)). You cant cheat the RNG gods. You have to pray to them. The Gods works in mysterious ways that we mortals cant comprehend. :P